The ai models & agents stories our desk judged worth your time, written so anyone can follow them.
Flagright's co-founder says AI should handle routine fraud and compliance alerts while people decide things like closing accounts, a shift that matters to anyone with a bank.
Read more: Fraud-Fighting Software Handles Routine Alerts, but People Make the Big CallsAndroid Auto reportedly keeps asking who to call even after Gemini hears the name, a bug that can make hands-free calling in the car frustrating.
Read more: Android Auto keeps asking who to call, even after you tell itUK shoppers now spend over £1 billion a year on paid AI tools, and Claude is gaining on ChatGPT. Here is what that means if you pay for one.
Read more: Britons now spend over £1 billion a year on AI subscriptionsHon Hai, which builds servers that use Nvidia chips, says October sales rose 46%. It suggests the rush to build AI computing power is still going strong.
Read more: Hon Hai's October sales jump 46% on strong demand for AI serversOpenAI's president says the first half of 2026 was rough, but sales jumped in the third quarter, helped by coding tools and work products.
Read more: OpenAI says a rough first half of 2026 gave way to a sales jumpMeta is reportedly dropping fruit codenames like Avocado, Mango and Watermelon for its AI projects, though the company has not confirmed the change.
Read more: Meta plans to stop naming its AI projects after fruitOpenAI's new image model, Images 2.5, cuts generation time by up to 50% and holds up better when you make repeated edits to the same picture.
Read more: ChatGPT's new image tool is up to 50% faster and edits betterEnkio now lets you turn a song and a few images into a music video from a chat in ChatGPT or Codex, using the credits you already have with Enkio.
Read more: You can now make a music video from inside ChatGPTAnthropic has committed $895.4 billion through 2045 to computing power, and your Claude limits and prices may depend on whether it can get enough.
Read more: Anthropic is spending nearly $900 billion to secure computing power for ClaudeA test sent an honest shopping robot to 200 US and UK stores. One in seven turned it away, and security software looks like the cause.
Read more: One in seven online stores turned away a shopping robot in a testAttackers chained a Chrome bug with a Snapdragon chip flaw to take over Android phones with no tap needed. Updating now closes the door.
Read more: Hackers broke into Android phones through Chrome without a single tapThe Financial Times says a new UK tax break for start-up founders is too small to matter much, and may only help if the government keeps it in place.
Read more: A new UK tax break for start-up founders may be too small to matterA new ElevenLabs guide says most transcription errors trace to three causes, and many can be fixed with small changes to recording habits and settings.
Read more: Why voice transcripts get words wrong, and how to fix itA Twilio post warns that a popular score for customer service AI counts people who give up and hang up as handled. It names six better measures.
Read more: Customer service AI can look like a success while customers are unhappyApple is reportedly trimming the daily AI spending allowance for its engineers, a sign that even big companies are checking whether AI pays off.
Read more: Apple is reportedly lowering how much its engineers can spend on AICrediMax and Mastercard will help Bahrain's stores sell to AI shopping assistants that search, compare and pay for shoppers within limits the shopper sets.
Read more: Bahrain Stores Get Ready to Sell to AI Shopping AssistantsCat Wu, who led Anthropic's Claude Code, is joining rival OpenAI, a sign that the fight over AI coding tools is getting more intense.
Read more: OpenAI hires the product leader behind Anthropic's Claude CodeAnthropic, maker of the Claude chatbot, has paid a news publisher for the first time. The AP deal could make AI answers about breaking news more reliable.
Read more: Anthropic pays a news publisher for the first time, striking a deal with APAn Anthropic engineer says AI writes code faster than people can learn it, and suggests a quiz before code is accepted. Here is why owners should care.
Read more: AI writes code fast, but teams may not understand what it wroteExpedia's CEO says AI assistants are good at helping people plan trips, but travelers still prefer to book on a brand they trust.
Read more: Travelers plan trips with AI helpers but still book on ExpediaNIQ and Similarweb are building a tool to show whether AI shopping chats lead to real sales. A first version is due in Q4 2026, so sellers have time to prepare.
Read more: Two firms plan to track sales that start with AI shopping chatsAnthropic is telling investors its AI assistant Claude can design chips, ahead of a stock debut reportedly valued near $1.5 trillion. Here is what to weigh.
Read more: Anthropic pitches its AI as a chip designer before stock debutNetflix hired Darshan Kantak, a longtime ad tech executive, to fix delays in its ad business, which affects its cheaper plan with ads and the companies that buy those ads.
Read more: Netflix hires a veteran ad tech leader to speed up its ads businessElevenLabs says its AI voice agents now plug into NiCE, Genesys, Amazon Connect and Five9 phone systems, so businesses can automate routine calls without a rebuild.
Read more: ElevenLabs lets AI voice agents answer calls inside big company phone systems
